For marine surface vessel with input saturation, model uncertainties, and unknown disturbances, this article proposes a path-following control and collision avoidance method based on error-constrained event-triggered switching line-of-sight guidance. This method can autonomously switch between path-following and collision avoidance modes, constrain the position and heading errors to satisfy safe navigation in confined waters, and reduce the computational frequency of the controller and the mechanical wear of the actuator. In the path-following mode, we construct an asymmetric modified barrier Lyapunov function to constrain the position errors and use the relative threshold event-triggered mechanism to reduce the update frequency of the guidance heading angle. In the collision avoidance mode, static or dynamic obstacles are avoided through the response of collision avoidance radius. In the heading and velocity control design, asymmetric modified barrier Lyapunov function and event-triggered mechanism are also applied, so that the marine surface vessel can track the guided heading angle and meet the heading error-constrained while reducing the update frequency of the controller. The adaptive auxiliary systems are used to compensate for the input saturation, and radial basis function neural networks and adaptive laws are used to approximate the model uncertainties and composite disturbances. According to the Lyapunov stability theory, all signals are semi-globally uniformly ultimately bounded and the Zeno phenomenon is avoided. Finally, the comparative experiment shows the superiority of the designed control strategy.